The United States has cut off key semiconductor shipments to a leading Chinese company. China has heavily subsidized its telecom suppliers and has started to pour government money into a domestic chip industry. We seem to be headed into a new world order: Western companies will use U.S. technology for semiconductors, and Chinese companies will use a separate Chinese semiconductor ecosystem.
This will be expensive. If each supplier can only sell in half of the world market, then the R&D investments by individual companies must shrink back in the future. That might be fine if R&D investments were stable or shrinking, as in other mature industries. But the R&D investment for semiconductors is growing! Developing a 5 nm process costs twice as much as 7 nm process development. And taping out a chip can cost more than $500 million for a major telecom product or computing platform.
This is the dark side of Moore’s law: We get better chips every 18 months, but the R&D investment also grows. For the past 30 years, this part has been no big deal, because globalization has created market growth that has made growing R&D cost unimportant. I believe that the new Cold Trade War will change the pattern. For the first time, the market may not grow fast enough to absorb rising R&D costs.
This will be true in the Western markets, where the R&D cost is spread among a large group of companies and therefore the investment is made by literally millions of investors. As the public markets start to understand the dire need for market growth, we are likely to see a reduction in the willingness of capital markets to support 3 nm or 2 nm or 1nm process development.
The impact could be even greater in China, where the end-user market is much smaller than the wider global market. China is making some gigantic investments in 5G networks right now, and they rival the rest of the world in some hand-picked markets where their government is willing to shovel money into development.
Specifically, Huawei may now be forced to develop a mainland Chinese foundry for its processor chips. The prime candidate, SMIC, is woefully behind in terms of performance in sub-18 nm process technology, and we find it doubtful that SMIC will be allowed to buy advanced equipment from KLA, LAM, Applied Materials, or the other key American sources. This means that—if we stay on our current track—China will first need to create technology for advanced fabrication equipment (at least $15B investment). Then they would need to develop the advanced semiconductor process (another $10B). They’ll need advanced chip design tools to replace Synopsys, Cadence, and others (another $5B+). And finally, they will need to re-create chip platforms such as Intel’s x86, nVIDIA GPUs, and 5G telecom chips ($20B or more, depending what you count). The overall expense would be $50B or more to re-invent the wheel that is driving the world markets right now.
But time is a factor as well. As China spends their $50B re-inventing the wheel, Yankee ingenuity will continue and the Americans will push on to new heights. After talking with key equipment companies, chip suppliers, and OEMs, we believe that China will need another five years to reach respectable semiconductor performance that is completely independent of the USA. Where will Analog Devices, Qualcomm and Intel be in 2025?
In a "hot" war, one side may win a strategic victory, but none of the soldiers on the battlefield win anything. The same is true in a Cold Trade War… it’s bad for all of the players on the ground. American suppliers could lose 30-40% of their market. (Or maybe more, since Huawei will push into APAC, Latin America and Africa to strive for a larger share of the world market). The impact to U.S. suppliers will be a loss of 4 percentage points of gross margin and 8-10 percentage points in operating profit, which means that R&D spending will drop and innovation will slow down.
On the other side of the Pacific, Chinese executives will have even bigger problems. They will lose 60-70% of their market, making a moonshot to "catch up" on technology impossible without massive government funding. The only way to achieve sustainability would be for a massive government program to artificially create the right technology for the 2025 market, including every step from sand to screen. Personally, I think that this kind of moonshot program is likely to fail.
Is it good policy for the USA to force the Chinese into bankruptcy, as we did to the Russians? That’s above my pay grade. As an analyst, I can report this analysis: A Cold Trade War with China in tech will cost U.S.-based semiconductor companies about $50B in profits per year over the next five years… a total of about $250B. It will cost China $95B per year in lost profit, plus $50B or more to recreate existing technologies… let’s call it at least $500B over the next five years, not including massive technology risk.
Is the American cost of $250B worthwhile to stop Chinese theft of technology? That’s up to Mr. Trump and Ms. Pelosi to decide. Is the Chinese cost of $500B+ worth paying, to avoid changes to IP law inside China? That’s for Mr. Xi to decide.
Joe Madden is principal analyst at Mobile Experts, a network of market and technology experts that analyze wireless markets. The team provides detailed research on small cell, base station, carrier Wi-Fi, and IoT markets. Madden currently focuses on trends in 5G, IoT, and enterprise markets for wireless infrastructure. Over 26 years in mobile communications, he accurately predicted the rise of digital predistortion, remote radio heads, small cells, and a mobile IT market. He validates his ideas with mobile and cable operators, as well as semiconductor suppliers, to find the match between business models and technology. Madden holds a physics degree from UCLA. Despite learning about economics at Stanford, he still obeys the laws of physics.
"Industry Voices" are opinion columns written by outside contributors—often industry experts or analysts—who are invited to the conversation by FierceWireless staff. They do not represent the opinions of FierceWireless.